The frequent clinical observation that rapists constitute a markedly heterogeneous group has led numerous clinicians and researchers to propose taxonomic systems aimed at identifying more homogeneous subgroups (7). Because all decisions about such offenders inevitably involve the interpretation of a particular case as a member of a larger group, the creation of a reliable and valid classification system could have important implications for treatment, management. and disposition. In spite of the potential importance of taxonomic considerations and their prominence in clinical theorizing, there have been a paucity of empirical studies of the various proposed systems. During the last decade we have been studying the critical taxonomic problems of rapists. We revised and operationalized a rationally derived system that incorporated the major types of rapists described in the clinical literature. In a series of studies (21- 24) we assessed the reliability and validity of this system, determining its strengths and weaknesses and identifying problem areas that required revision. We also cluster analyzed samples of rapists, and tested the validity of various types that evidenced cross-sample stability (e.g.,25). Finally, we integrated these two sets of studies and created anew system. This new system solves all of the problems we identified in our analyses of our original rationally derived topology, incorporates the new stable types that emerged in our cluster analytic studies, and provides anew empirically-based structure that has substantial heuristic potential. The major aim of the current proposal is to test the reliability and concurrent and predictive validity of this new system. Because of the extensive computerized clinical and criminal file data base that we have established on a large sample of rapists and the post release follow-up data we have on a subsample of these offenders, a major validation study can be conducted efficiently in a relatively short time.